New table estimates probability of finding PCa

February 1, 2005

New York--Factors traditionally linked to risk assessment of prostate cancer—PSA, digital rectal examination, and age—are summed up in a new table that is designed for ease of use.

New York-Factors traditionally linked to risk assessment of prostate cancer-PSA, digital rectal examination, and age-are summed up in a new table that is designed for ease of use.

The table is a boiled-down, cross-referenced risk-stratification system that a team of researchers from Colorado (E. David Crawford, MD); Michigan (Mani Menon, MD); Tyrol, Austria (Georg Bartsch, MD); and New York (Ashutosh Tewari, MD) have compiled to help any office-based physician decide whether transrectal ultrasound-guided prostate biopsy is warranted in a patient who has undergone initial prostate cancer screening.

"This will help physicians in individualizing biopsy strategies based on these parameters," said Dr. Tewari, associate professor of urology and director of robotic prostatectomy and urologic outcomes at Weill Medical College of Cornell University, New York.

Validated with two data sets The analysis was made by utilizing a large screening database to derive risk estimates, particularly the probability of finding cancer by transrectal ultrasound-guided biopsy.

The table was developed based on data from the Tyrol Screening Project, a European study co-authored by Dr. Bartsch that involved 3,814 men. Variables used in the model were age, DRE findings, and serum PSA level. Validation was accomplished by means of a step-wise logistic regression model. A 95% confidence interval was calculated for each mean-probability estimate to ensure accuracy. In fact, the tables were validated with two data sets from two different institutions-Virginia Mason Medical Center and Stanford University Medical Center, Dr. Tewari said.

In the Tyrol studies, the mean age of the men was 63 years, and mean PSA was 9.6 ng/mL. DRE was suspicious in 581 of the men (15%). Carcinoma was confirmed by biopsy in 1,022 (nearly 27%).

At 90% sensitivity, the model had a specificity of 12% and 10% for Virginia Mason and Stanford University validations, respectively.

Recently, Dr. Tewari and his group have updated the dataset to include more than 7,000 patients and have added additional variables such as free and complex PSA and TRUS volume estimates. An additional model has been constructed that not only predicts possibility of cancer in the biopsy but also predicts the likelihood of finding aggressive cancer (Gleason score 7 and higher).

The end result is a guide that shows the estimated probability of a positive TRUS-guided prostate biopsy based on the three parameters.

"Such a table can be used by practitioners and patients when assessing diagnostic options for suspected cancer," Dr. Tewari said.

The tables can be obtained by contacting Dr. Tewari at AshTewari@med.cornell.edu
.